Physicalists are committed to holding that the physical necessitates everything else which is the case. If physicalism is true of our world, the physical nature of our world fully determines where the shopping centres are and which ones are the biggest, what you and I are feeling and thinking, the current rate of inflation, and so on. The details are controversial but the basic idea is not.
What is controversial - in principle and not just detail of formulation - is whether physicalists are committed to holding that some suitably rich, true account of the physical way things are a priori entails the psychological, political, social, weather, etc., way things are. Are physicalists committed to what I call the a priori passage principle, the view that for each true statement concerning our world, there is a statement in physical terms that a priori entails that statement? 1
Deniers of a priori passage point out following Kripke (1980) and Putnam (1975) that there are conditionals of the form 'if things are thus and so H2O-wise, then things are such and such water-wise', and 'if things are thus and so molecular kinetic energy-wise, then things are such and such heat-wise in gases', which are necessary a posteriori truths that go from truth to truth but which are not a priori.
I think the example of H2O and water is a bad one for the deniers of a priori passage. My argument (in Jackson 1992; 1994; 1998:80-3) for this conclusion has a simple structure. I take a putative inference from some truth about how things are framed in terms of H2O to something about how things are framed in terms of water, which is valid in the sense of being necessarily truth-preserving but which is not valid in the sense of being a priori. I then provide an additional, true, contingent, a posteriori premise about how things are framed in terms of H2O, which, I argue, makes the inference valid in the a priori sense. Although I turn the trick for only a very small number of cases, once you have seen one, you have seen them all. It is obvious how to generalize to other necessarily truth-preserving inferences from H2O to water, and how to extend to similar examples involving gold, heat, etc. and, indeed, all the examples that arise from the standard examples of necessary a posteriori